Generative and Agentic AI Architect

Project overview

Within a department that combines innovation, research and development, and technological expertise, you will join the Architecture and Cloud team. This team supports application projects with their architectural choices, design, and continuous integration, in containerized environments and private or public cloud infrastructures.

You will define the architectural framework for generative and agentic AI solutions, then support their production deployment. You will contribute to model and platform choices, with particular attention to security, compliance, performance, and cost control.

Responsibilities

You will also contribute to the integration of agentic AI into development practices. As such, you will be expected to:

Profile

You have at least nine years of experience in software development and architecture, including a minimum of three years on AI or generative AI projects. You are proficient in Python, especially FastAPI, and know Java, particularly Quarkus and LangChain4j.

Expected Expertise

• Define the architecture for generative and agentic AI use cases: assistants, RAG systems, agents, and multi-agent architectures; design reusable models.
• Monitor market developments, recommend suitable solutions, and share best practices within the AI community.
• Participate in the design and evolution of the AI platform: provision of LLMs, model gateways, vector databases, and agent orchestration, particularly on OpenShift and in the public cloud.
• Integrate security and governance requirements: prompt injection and data leak prevention, authorization control, regulatory compliance, and responsible AI principles.
• Define methods for evaluating models and agents, monitor their quality and performance, and contribute to LLMOps observability and cost management.
• Produce architectural documents and technical specifications, participate in reviews, and coordinate work with IT and business teams.
• Support the transition of prototypes and pilots to production, in conjunction with infrastructure, security, and operations teams, and then contribute to operational maintenance.
• Design and prototype agents and workflows, and participate in the modular structuring of the platform.
• Connect agents to the information system's tools and data via MCP and internal APIs, within a secure and governed framework.
• Automate the deployment of models and agents with a GitOps approach, particularly via ArgoCD or OpenShift GitOps, and set up LLMOps pipelines: evaluation, prompt versioning, and monitoring.
• Support developers in using code assistants and act as an AI referent within the team.
• Provide technical support to projects and their teams.
• Generative and agentic AI: LLMs (GPT, Claude, Gemini, Mistral, Llama), prompt engineering, RAG, and frameworks such as LangChain, LangGraph, LlamaIndex, CrewAI, AutoGen, or Google ADK; MCP and A2A protocols.
• AI Platforms: Vertex AI, Azure OpenAI, and/or Amazon Bedrock; deployment of open-source models in private cloud with OpenShift AI, vLLM, KServe, or Ollama; model gateways like LiteLLM; understanding of GPU constraints.
• Data and Search: vector databases such as pgvector, Milvus, Qdrant, and OpenSearch/Elasticsearch; chunking, embeddings, hybrid search, and reranking; SQL and PostgreSQL.
• LLMOps and Evaluation: evaluation tools like Ragas or DeepEval; observability and traceability with Langfuse, OpenTelemetry, or MLflow; guardrails, prompt management, and FinOps monitoring.
• Containers and DevOps: Docker, Kubernetes, OpenShift, CI/CD, Git, and Jenkins; GitOps practices with ArgoCD; advanced use of code assistants such as Claude Code or GitHub Copilot.
• Architecture, Security, and Compliance: cloud infrastructure components (networking, storage, reverse proxy, TLS certificates), AI application security (OWASP Top 10 for LLM, secrets management, RBAC), AI Act, and responsible AI.

Project details

• Start: flexible
• Remote: hybrid
• Location: Paris, France, FR
• Rate: 75-75 EUR/h
• Languages: French